INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

INTIMAL HYPERPLASIA, characterized by smooth muscle cell (SMC) accumulation and matrix deposition, often develops after angioplasty or bypass grafting, and compromises the long-term patency of these interventions. The cause of anastomotic intimal hyperplasia is not known precisely, but mitogens released from platelets, leukocytes, endothelial cells (EC), and SMC may contribute to the initiation and perpetuation of the processes. SMC in intimal hyperplastic lesions of grafts are in a synthetic phenotype and produce high levels of platelet-derived growth factor (PDGF) (3, 15, 18, 23). Factors controlling SMC production of PDGF are unknown, but oxidized low-density lipoproteins (oxLDL) may play a role.

Elevated cholesterol is a well-recognized risk factor for angioplasty or vascular graft failure, just as it is for atherosclerosis. Oxidatively modified lipids and lipoproteins may play a role in vascular graft failure because lipid oxidation products accumulate in prosthetic grafts, and oxLDLs possess a number of properties that would adversely affect graft healing, including stimulation of SMC proliferation and extracellular matrix production. Our previous studies show that Dacron graft material can stimulate monocytic cells to oxidize LDL in vitro (29) and that products of lipid oxidation accumulate in prosthetic grafts in vivo (28). The exact relationship between accumulation of lipid oxidation products and the development of anastomotic intimal hyperplasia is currently unknown.

SMC cultured from prosthetic grafts produce higher levels of PDGF and collagen than do aortic SMC (19, 23). These phenotypic differences are maintained for several passages in culture. OxLDL is reported to stimulate PDGF A-chain transcripts in late passage human arterial SMC (26), so the ability of oxLDL to stimulate PDGF-AA production by graft and aortic SMC was studied. Graft SMC not only produced more PDGF than aortic SMC under basal conditions, but the graft SMC responded to oxLDL, superoxide generation, and H₂O₂ with a significantly greater increase in PDGF production than aortic SMC. N-acetylcysteine (NAC) blocked the stimulation of PDGF production by oxLDL. The elevated PDGF production may contribute to the development of anastomotic intimal hyperplasia by stimulating SMC migration and proliferation that eventually may cause synthetic vascular graft occlusion.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Graft implantation and removal. To evaluate PDGF production by SMC from the aorta and prosthetic vascular grafts, adult female mongrel dogs underwent placement of Cooley double-velour knitted Dacron thoracoabdominal grafts (Meadox Medicals). The grafts (8 mm internal diameter and 22 cm long) were implanted following the method described by Graham et al. (9). After 16-26 wk, the grafts and native vessels were carefully isolated from surrounding tissue and flushed with medium 199 (M199) tissue culture medium (Sigma). Tissue on the outer surface of the graft was removed during the harvest. The protocol for animal studies was approved by the institutional committee on animal use. All procedures and care complied with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, 1996).

SMC harvest and culture. Cells were harvested from the proximal aorta (ascending aorta and aortic arch) and the distal abdominal aorta below the graft anastomosis. In addition, cells were isolated from midgraft sections. The luminal surface was incubated in collagenase A (630 U/ml; Boehringer-Mannheim) for 10 min at 37°C in a 5% CO₂ atmosphere, and EC were removed mechanically. The medial layer of the aorta was then stripped and the media and graft divided into 2 × 2 mm² pieces. These pieces were incubated for 8 h at room temperature with continuous gentle shaking in M199 that contained 15 U/ml elastase type III (Sigma), 170 U/ml collagenase A, 2 mg/ml crystalline bovine serum albumin (Boehringer Mannheim), and 0.375 mg/ml soybean trypsin inhibitor (Boehringer Mannheim). The SMC were collected with the use of a cell strainer (model HX86695, Falcon), resuspended in M199, pelleted by centrifugation, plated in serum-coated tissue culture plates, and grown to confluence in M199 with 20% fetal bovine serum (FBS; Hyclone Laboratories), penicillin (100 U/ml), and streptomycin (100 µg/ml). For all experiments, passage 2 or 3 SMC were plated in 24-well culture plates at 30,000 cells/cm² and grown to confluence in M199 containing 20% FBS at 37°C in a 5% CO₂ atmosphere and then changed to medium containing 5% FBS and the compound to be tested. The identity and homogeneity of the SMC population were confirmed immunohistochemically with antibody to SMC -actin (Sigma).

Enzyme-linked immunosorption assay for canine PDGF-AA. A heterogeneous sandwich assay was developed for quantifying PDGF-AA production by canine SMC. Briefly, Nunc-immuno 96-well enzyme-linked immunosorbent assay (ELISA) plates (MaxiSorp; Naperville, IL) were coated with 50 µl of affinity purified polyclonal goat anti-human PDGF-AA antibody (no. AD-221, R&D) at a dilution of 0.5 µg/ml of coating buffer (0.6 M NaCl, 0.26 M H₃BO₄, and 0.08 N NaOH, pH 9.6) for 18 h at 4°C. After excess capture antibody was removed and blocked with 3% bovine serum albumin and phosphate-buffered saline for 120 min at 25°C, 100-µl aliquots of recombinant human PDGF-AA (UBI) or samples were added and incubated for 18 h at 4°C. Detection antibody, polyclonal rabbit antihuman PDGF-AA (Research Diagnostic), and 0.5 µg/ml blocking buffer was added for 2 h at room temperature. Biotinylated anti-rabbit IgG (BioGenex) (0.1 µg/well) was then added for 1 h. After being washed, 0.1 µg/well streptaviden peroxidase (Sigma) was added for 45 min at room temperature in the dark. o-Phenylenediamine (OPD) chromogenic substrate (DAKO) (100 µl/well) was added for 20 min, and the reaction was terminated with the addition of 100 µl/well 0.5 M H₂SO₄. The plates were read on an ELISA plate scanner (model ELX 808, Bio-Tek Instruments) at 492 nm. The detection limit for PDGF-AA was 20 pg/ml. ELISA was validated for canine PDGF by measurement of serial dilutions from 1:1 to 1:16 of a concentrated sample of conditioned medium. The curve that was generated paralleled the standard, demonstrating reliable detection of canine PDGF-AA over a range of 30-2,000 pg/ml (Fig. 1).

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Enzyme-linked immunosorbent assay (ELISA) for canine platelet-derived growth factor (PDGF). A: standard curve was created using known amounts of human PDGF-AA (n = 3). B: sample of conditioned medium from graft smooth muscle cells (SMC) (n = 5) was concentrated, serial dilutions were then made from 1:1 to 1:16, and PDGF was measured by ELISA. The curve generated paralleled the standard, demonstrating reliable detection of canine PDGF-AA in the range of 30-2,000 pg/ml.

Protein synthesis assay. To measure protein synthesis, SMCs were incubated with [³H]leucine (0.5 mCi/ml, DuPont-NEN) and then harvested. [³H]Leucine that was incorporated into trichloroacetic acid (TCA; Sigma)-precipitable material was measured, as described by Fox and DiCorleto (8). Briefly, to measure secreted protein synthesis, the conditioned medium was collected and protein precipitated using chilled 5% TCA and 0.25% tannic acid. Cell-associated protein synthesis was determined by lysing the remaining cells and by precipitating protein with chilled 5% TCA-tannic acid solution. In both cases, the precipitate was solubilized with 0.5 N NaOH, neutralized with 50 µl 6 N HCl, and the radioactivity in each sample was then determined with a liquid-scintillation counter. Results were expressed as disintegrations per min per microgram of DNA.

DNA measurement. After the conditioned media were removed from the wells, cells were harvested and DNA quantitated using the method described by Labarca and Paigen (16). Briefly, cells were harvested by incubation in 0.05% trypsin solution and then lysed with a sonicator. The cell lysate (25 µl) was then diluted with phosphate/NaCl-buffered solution (4 M NaCl) and an equal volume of bis-benzimide solution (Sigma) added. Samples (200 µl) and DNA standards prepared with stock calf thymus DNA (Sigma) were added to 96-well plates (no. HX86610C, Costar) in triplicate. Plates were read on a Microplate Fluorescence Reader (model FL600, Bio-Tek Instruments) with excitation set at 360 nm and emission at 460 nm. The DNA standard curve was prepared and results were fit by nonlinear regression using the logistic equation.

LDL isolation and oxidation. Human LDL (density = 1.019-1.063 g/ml) was prepared by sequential ultracentrifugation of fresh, citrated plasma to which EDTA was added before centrifugation (11). Endotoxin was measured using a commercially available kit (Sigma). Cholesterol was measured using cholesterol calibrators (Sigma). Purified LDL was filtered through a 0.45-µm filter. LDL was stored in the dark at 4°C under nitrogen for no longer than 3 wk before use. Immediately before use, LDL was dialyzed at 4°C for 48 h against saline to remove the EDTA. For oxidization, LDL was diluted to 10 mg/ml in phosphate-buffered saline and incubated with 1.5 µM copper sulfate for ~18 h at 37°C until the clear solution became colorless. The level of LDL oxidation was monitored by the formation of thiobarbituric acid reacting substances (TBARS), measured by the method of Schuh and colleagues (25). Samples were analyzed by fluorimeter with excitation at 515 and emission at 553. TBARS were expressed as nanomoles of malondialdehyde (MDA) per milligram of cholesterol. The TBARS of oxLDL used in the experiments ranged from 5 to 8 nmol MDA/mg cholesterol, whereas TBARS for the native LDL ranged from 0.3 to 0.5 nmol MDA/mg cholesterol.

Reverse transcriptase-polymerase chain reaction. Reverse transcriptase-polymerase chain reaction (RT-PCR) was used to amplify the PDGF A-chain and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA isolated from the canine SMC. The primers used to amplify the PDGF A-chain were chosen based on previously published sequences (30). For canine GAPDH primers and probe were designed using a software program (Premier Biosoft International) and the cDNA sequence downloaded from a NCBI database (10). Oligonucleotides were synthesized commercially (Operon Technologies). The oligonucleotide sequences used for PDGF A-chain were as follows: sense primer (5'-CCT-GCC-CAT-TCG-GAG-GAA-GAG-3'), anti-sense primer (5'-TTG-GCC-ACC-TTG-ACG-CTG-CG-3'), and probe (5'-CGT-GCG-TGG-AGG-TGA-AAC-G-3') (30, 32). The corresponding sequences used for canine GAPDH were as follows: sense primer (5'-CAT-GTT-TGT-GAT-GGG-CGT-GAA-C-3'), anti-sense primer (5'-GTG-GCA-GTG-ATG-GCA-TGG-AC-3'), and probe (5'-TGG-ATG-ACT-TTG-CTA-GAG-G-3').

PCR-ELISA. After confirmation of the RT-PCR products as above, digoxigenin-labeled products were generated in subsequent studies. Ten microliters of 200 µM digoxigenin-labeled dNTP (Roche Molecular Biochemicals) was added to the PCR reaction mixture, RT-PCR was performed as described above, and products were detected by PCR-ELISA using a commercially available digoxigenin detection kit (Roche Molecular Biochemicals). Briefly, a digoxigenin-labeled amplified product was denatured for 10 min, and appropriate biotinylated probe was added in 500 µl of hybridization buffer. For negative controls, DNA was omitted or the inappropriate probe was used. Aliquots of 200 µl were added to duplicate wells of a streptavidin-coated 96-well plate and incubated for 4 h at 42°C. A bound product was detected with peroxidase-labeled anti-digoxigenin-peroxidase-conjugated antibody by standard colorimetric reaction with the use of the peroxidase substrate kit ABTS. An optical density at 405 nm was measured using a microplate reader. PDGF A-chain and GAPDH mRNA were amplified in parallel. Results were expressed as a ratio of the optical density for PDGF A-chain mRNA to the optical density for GAPDH. Northern blot analysis verified that GAPDH gene expression did not change during SMC incubation in oxLDL.

Statistics. All data represent means ± SE. Experiments were performed in triplicate with at least three different cell isolates. Data evaluation was performed by analysis of variance with the use of InStat (GraphPad Software) or t-test using Statview (BrainPower). Differences were considered statistically significant at P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In the present study, we explored the effect of oxLDL on PDGF-AA secretion with aortic and graft SMC. Constitutive secretion of PDGF-AA by graft SMC over 72 h was significantly higher than that by aortic SMC at 78 ± 2 and 21 ± 3 pg PDGF/µg DNA, respectively, P < 0.001 (Fig. 2). OxLDL (TBARS range of 5-8 nmol MDA/mg cholesterol) induced a concentration-dependent increase in PDGF-AA production by graft SMC, which was significantly greater than aortic SMC. PDGF production was maximum at a concentration of 300 µg cholesterol/ml, being 248 ± 21 and 37 ± 6 pg PDGF/µg DNA for graft and aortic SMC, respectively. Native LDL (TBARS range of 0.3-0.5 nmol MDA/mg cholesterol) had no effect on graft or aortic SMC PDGF production. The concentrations of oxLDL studied were not toxic in 72 h as shown by stable levels of DNA and protein synthesis. OxLDL induced a statistically significant increase in PDGF-AA production by graft SMC (P < 0.001) but not by aortic SMC.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Effect of oxidized low-density lipoprotein (oxLDL) and native LDL on PDGF-AA production by graft and aortic SMC. Passage 2 graft and aortic SMC (n = 10) were grown to near confluence in medium 199 (M199) with 20% fetal bovine serum (FBS) and then changed to M199 with 5% FBS and varying concentrations of LDL or oxLDL. After 72 h, PDGF in the conditioned medium was measured by ELISA, and the cells were harvested for DNA measurement. PDGF-AA levels were standardized per microgram of DNA. Data are means ± SE. *P < 0.001 vs. graft SMC no oxLDL; **P < 0.01 vs. no oxLDL; ***P < 0.001 vs. no oxLDL.

OxLDL is known to produce an oxidative stress that might be responsible for the increase in PDGF production. To determine the effect of other oxidative stresses, PDGF production was measured after incubation of graft and aortic SMC with H₂O₂ or 6-anilinoquinoline-5,8-quinone (LY-83583; Sigma), a substance known to induce superoxide production by SMC (1). As with studies using oxLDL, graft SMC were found to be much more responsive to the stimulatory effects of LY-83583 or H₂O₂ than were the aortic SMC. PDGF production by graft SMC increased from 66 ± 4 to 218 ± 6 pg PDGF/µg DNA (P < 0.001) as the LY-83583 concentration was increased from 0 to 1.2 µM (Fig. 3). Over the same concentration range, PDGF production by aortic SMC production rose from 23 ± 2 to 37 ± 3 pg PDGF/µg DNA (P > 0.05). Similarly, as the H₂O₂ concentration was increased from 0 to 50 µM, PDGF production by graft SMC increased significantly from 73 ± 3 to 201 ± 9 pg PDGF/µg DNA (P < 0.001), but PDGF production by aortic SMC was relatively constant, changing from 19 ± 2 to 24 ± 2 pg PDGF/µg DNA (P > 0.05) over the same H₂O₂ concentration range (Fig. 4). At concentrations >50 µM, decreases in DNA were noted in some experiments, suggesting toxicity. These findings suggest that oxygen free radicals cause an increase in SMC production of PDGF-AA and that graft SMC are more responsive to the oxidative stress than aortic SMC.

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Effect of 6-anilinoquinoline-5,8-quinone (LY-83583) on PDGF-AA production by graft and aortic SMC. Graft and aortic SMC (n = 5) were grown to near confluence in M199 with 20% FBS and then changed to M199 with 5% FBS and varying concentrations of LY-83583. After 72 h, PDGF in the conditioned medium was measured by ELISA, and the cells were harvested for DNA measurement. PDGF-AA levels were standardized per microgram of DNA. Data are means ± SE. *P < 0.05 vs. no LY-83583; **P < 0.001 vs. no LY-83583.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Effect of H2O2 on PDGF-AA secretion by graft and aortic SMC. Graft and aortic SMC (n = 6) were grown to near confluence in M199 with 20% FBS and then changed to M199 with 5% FBS and varying concentrations of H2O2. After 72 h, PDGF in the conditioned medium was measured by ELISA, and the cells were harvested for DNA measurement. PDGF-AA levels were standardized per microgram of DNA. Data are means ± SE. *P < 0.001 vs. no H2O2.

The time course of the increase in graft SMC production of PDGF was investigated. PDGF production in response to oxLDL, H₂O₂, and LY-83583 was assessed at 24, 48, and 72 h. OxLDL, H₂O₂ and LY-83583 stimulated an increase in PDGF-AA production compared with control, and the increase was linear over time (Fig. 5).

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Time course of PDGF-AA production by graft SMC. Graft SMC (n = 5) were grown to near confluence then 300 µg/ml oxLDL, 50 µM H2O2, or 1.2 µM LY-83583 was added. Conditioned medium and cells were collected at 24, 48, or 72 h for measurement of PDGF-AA by ELISA and DNA by fluorimetric assay. PDGF-AA levels were standardized per µg DNA. Depicted in the graph is the PDGF secretion in the presence of each agonist minus the PDGF produced under control conditions during the same time.

Protein synthesis by graft and aortic SMC was measured to assess cell viability and determine the specificity of changes in PDGF secretion after incubation with agonists. SMC protein synthesis was determined in conditions identical to those for PDGF assay. Confluent aortic and graft SMC were incubated with oxLDL (300 µg/ml), H₂O₂ (50 µM), or LY-83583 (1.5 µM). [³H]leucine (0.5 mCi/ml) was added for determination of protein synthesis. Both secreted and cell-associated protein synthesis was constant in control conditions and with all agonists studied (Fig. 6). This suggested that the stimulatory effect of oxLDL, H₂O₂, and LY-83583 was relatively specific for PDGF, and the increase in PDGF was not simply part of a generalized increase in cellular protein synthesis. In addition, agonists were not toxic in the concentrations used as demonstrated by protein synthesis and visual morphologic analysis of SMC.

View larger version (17K): [in this window] [in a new window]   Fig. 6.   Effect of LDL, OxLDL, LY-83583, and H2O2 on protein production by graft SMC. Graft SMC (n = 3) were harvested, passaged, plated, and grown to near confluence using identical methods as for measurement of PDGF production. [3H]leucine and various agonists, including oxLDL (300 µg/ml), LY-83583 (0.5 µM or 1.2 µM), or H2O2 (50 µM) were added for 72 h. [3H]leucine incorporation into trichloroacetic acid (TCA)-precipitable material in the conditioned medium and in the cell layer was measured and reported as disintegrations per minuter per microgram of DNA. Total protein synthesis was not significantly altered by any treatment.

The studies with oxLDL, H₂O₂, and LY-83583 suggest that reactive oxygen species (ROS) may alter PDGF production. Therefore, the ability of antioxidants to prevent the increase in PDGF production by graft SMC was evaluated. Several antioxidants, including superoxide dismutase (SOD; 500 U/ml), catalase (4,000 U/ml), and NAC (1 mM) were tested for their effect on oxLDL-mediated increase in PDGF-AA production by graft SMC. SOD and catalase had no effect, even when linked to polyethylene glycol (data not shown). This may have been due to their relatively short duration of action. NAC blocked the increase in PDGF secretion induced by oxLDL (Fig. 7). NAC was effective for concentrations of oxLDL up to 400 µg cholesterol/ml. The ability of an intracellular antioxidant to inhibit the increase in PDGF production suggests that the oxidative stress created by oxLDL is at least in part responsible for its effect on PDGF secretion.

View larger version (16K): [in this window] [in a new window]   Fig. 7.   Inhibition of oxLDL-induced PDGF-AA secretion by N-acetylcysteine (NAC). Graft SMC (n = 5) were grown to near confluence and then incubated with oxLDL (200, 300, or 400 µg/ml) in the presence and absence of 1 mM NAC. After 72 h, PDGF in the conditioned medium was measured by ELISA, and the cells were harvested for DNA measurement. PDGF-AA levels were standardized per microgram of DNA. Data are means ± SE. *P < 0.001 vs. oxLDL with no NAC.

The mechanism by which PDGF-AA production was increased was explored by determining PDGF gene expression. Graft and aortic SMC were grown to confluence and agonists added for 72 h. The conditioned medium was removed for measurement of PDGF, and the cells were harvested for isolation of mRNA. PDGF gene expression was determined using RT-PCR and represented as the ratio of PDGF A-chain signal to GAPDH signal (Fig. 8). PDGF-AA production and mRNA expression were increased in parallel by oxLDL, H₂O₂, and LY-83583. OxLDL, H₂O₂, and LY-83583 induced greater increases in PDGF A-chain mRNA expression in graft SMC than in aortic cells. NAC inhibited the oxLDL-induced increase in mRNA in graft SMC.

View larger version (14K): [in this window] [in a new window]   Fig. 8.   Effect of LDL, OxLDL, LY-83583, and H2O2 on PDGF A-chain gene expression in graft and aortic SMC. Graft and aortic SMC (n = 3) were harvested, passaged, and grown to near confluence using identical methods as for measurement of PDGF production. Various agonists, including native LDL (300 µg/ml), oxLDL (300 µg/ml), LY-83583 (1.2 µM), or H2O2 (50 µM), were added. In addition, the effect of pretreatment with NAC (1 mM) on changes in gene expression induced by oxLDL (300 µg/ml) was also measured. After 72 h, SMC were harvested and mRNA isolated. Gene expression for PDGF A-chain and GAPDH was determined by reverse transcriptase-polymerase chain reaction (RT-PCR), and digoxigenin-labeled products were detected by ELISA. Results were expressed as the ratio of optical density for PDGF A-chain to the optical density for glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Representative results of one of the three separate experiments are shown. Data are means ± SE.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Intimal hyperplasia, which develops at the anastomoses of prosthetic vascular grafts and contributes to their failure, has many parallels to atherosclerosis. Graft implantation is accompanied by monocyte invasion, accumulation of lipid oxidation products, limited EC migration, and SMC proliferation (5, 28). This similarity in cellular dysfunction between vascular graft healing and atherosclerosis suggests that common factors may contribute to atherogenesis and graft failure. The lipid oxidation hypothesis of atherogenesis suggests that the cellular dysfunction occurs secondary to accumulation of oxidized lipids. Elevated cholesterol, a well-recognized risk factor for atherosclerosis, contributes to vascular graft failure (2, 4, 13). Interestingly, LDL is deposited preferentially at anastomoses of polytetrafluoroethylene grafts in rabbits (2, 20), the same regions that are susceptible to intimal thickening due to SMC proliferation and extracellular matrix accumulation. Lipid oxidation products that accumulate in grafts may contribute to the development of intimal hyperplasia by inducing changes in cellular function that would promote SMC accumulation.

OxLDL stimulates SMC proliferation and migration, and one mechanism may involve increased PDGF production. Although previous studies have suggested that oxLDL stimulates PDGF secretion (26, 33), our results show that the constitutive production of PDGF by SMC, as well as PDGF production in response to oxLDL, H₂O₂, or LY-83583, is dependent on the phenotype of the SMC. Contractile SMC from normal arteries produce a little PDGF, but modified SMC from atherosclerotic lesions and prosthetic grafts exhibit a synthetic phenotype and produce elevated levels of PDGF (14, 17, 23). OxLDL, LY-83583, and H₂O₂ have little effect on aortic SMC, but stimulate a marked increase in PDGF production by graft SMC. Zwijsen et al. (33) found that exposure of the human umbilical artery SMC to LDL increased PDGF gene expression, as determined by in situ hybridization. The effect was abrogated in the presence of BHT, suggesting that oxidation products were responsible, but the LDL oxidation that occurred during the 8-h incubation was not measured. Stiko-Rahm et al. (26) found that a 1- to 4-h exposure to Cu²⁺-oxidized LDL increased PDGF A-chain mRNA in late passage (passages 8-16) human arterial SMC and increased the expression of PDGF - and -receptors. In the aforementioned studies the SMC would have expressed a synthetic or undifferentiated phenotype, similar to graft SMC. Interestingly, we found that oxLDL had no significant effect on PDGF production by early passage (passage 2) aortic SMC that exhibited a more differentiated, contractile phenotype, but did stimulate PDGF production by late passage (>passage 6) aortic SMC after they had undergone phenotypic modulation in culture (data not shown). Changes in protein production paralleled changes in gene expression for PDGF A-chain. Thus oxLDL stimulated PDGF production by SMC, but only synthetic or dedifferentiated SMC, such as SMC in intimal hyperplastic or atherosclerotic lesions. The ability of oxLDL to promote a phenotypic transformation was not addressed in our study.

The mechanism by which oxLDL stimulates PDGF production may involve production of ROS. OxLDL induces superoxide production by SMC (12), and oxidative stress stimulates PDGF gene expression in EC in vitro and in lung in vivo (6, 22, 27). Furthermore, a lipid peroxidation product, 4-hydroxy-2-nonenal, induces PDGF-AA production and mitogenesis of rat SMC, the latter being inhibited by the antioxidant NAC (24). Our results support the role of ROS in the stimulation of PDGF production by oxLDL, because LY-83583, which stimulates superoxide production by NAD(P)H oxidases, and H₂O₂ increase PDGF production and gene expression by graft SMC. Furthermore, NAC abrogates the increase in PDGF production and gene expression induced by oxLDL. These findings support the premise that oxLDL stimulates PDGF production by causing an oxidative stress and increasing generation of ROS, which in turn increases PDGF gene expression and protein production. The stimulatory effect of oxLDL, LY-83583, and H₂O₂ on PDGF production by graft SMC but not aortic SMC suggests that graft SMC are more susceptible to oxidative stress. This could be the result of higher intrinsic ROS production by graft SMC or depleted antioxidant systems. West et al. (31) have shown that experimental vein grafts produce more superoxide than normal vein and that the principal source of the increased superoxide is NAD(P)H oxidase in the hyperplastic intimal SMC. This superoxide may be a mediator of graft intimal hyperplasia by stimulating production of PDGF and other mitogens, which in turn promote SMC migration and proliferation and contribute to intimal hyperplasia. Our results support this hypothesis, although our evidence for the role for free radicals is indirect, and the specific ROS potentially involved is not identified.

PDGF production in response to oxLDL could adversely impact on graft healing and contribute to the development of anastomotic intimal hyperplasia by stimulating SMC migration and proliferation. Ferns et al. (7) found that infusion of antibody to PDGF limited neointimal hyperplasia after rat carotid arterial injury by interfering with migration of medial SMC into the neointima. Previous studies in our laboratory showed that graft SMC produce high levels of PDGF (23) but suggested that PDGF production was not functioning in an autocrine fashion to stimulate SMC proliferation (21). Although graft SMC proliferated in response to PDGF, the response was blunted and fewer receptors were available for binding PDGF on graft SMC compared with aortic SMC. At the anastomosis, where lipid accumulation is the most concentrated, the PDGF-producing graft SMC are adjacent to PDGF-responsive arterial SMC. OxLDL-stimulated PDGF production may increase proliferation and migration of aortic SMC and contribute to the development of intimal hyperplasia in the anastomotic region. In addition, other factors, including the stimulation of extracellular matrix protein production by oxLDL, may promote the development of intimal lesions.

Our study shows that oxLDL consistently stimulates PDGF production by graft SMC that are in a secretory phenotype but has little effect on contractile aortic SMC production. Our observation that the origin (or phenotype) of the SMC is a determining factor in the response to oxLDL is a unique finding. Factors accounting for this observation, such as differences in antioxidant mechanisms or altered production of ROS by graft and aortic SMC, remain to be determined in future studies. An improved understanding of the effect of lipid oxidation products, which accumulate in prosthetic grafts, on the function of cells that migrate onto the grafts, will allow targeted interventions to improve cell function and graft patency.